The reproductive epidemiology project emphasizes the development and application of methods for measuring human reproductive damage. Reproductive damage can include infertility, sub-clinical early fetal loss, spontaneous abortion, impaired fetal growth, and low birthweight. Each of these outcomes can be produced by environmental factors, and each represents a possible endpoint for detecting the effects of toxins of human reproduction. This year we are completing a prospective study of very early pregnancy loss. By analyzing daily urine specimens from women who are trying to become pregnant, we are able to detect pregnancies that end before women know they are pregnant. Our preliminary estimate is that between 20% and 25% of pregnancies in this study end in unrecognized loss. Further work will be done to determine if the risk of such loss is related to common exposures such as use of alcohol, tobacco, caffeine beverages or medications. We used the same methods of urine analysis to explore the mechanism by which IUDs work as a contraceptive. There has been some evidence to suggest that the IUD does not actually prevent conception, but instead interfers with implantation. This possibility that IUDs prevent clinical pregnancies by causing early abortions was studied in a group of 40 women with IUDs. We looked for evidence of unrecognized pregnancy loss by testing daily urine specimens. We found only one such loss in more than 100 cycles. This provides the most conclusive evidence to date that the IUD's contraceptive action occurs before the time of implantation. In a separate analysis of the early pregnancy study data, we have examined the number of cycles that women required to become clinically pregnant. (Number of cycles to pregnancy is a function of the couple's monthly probability of conception). We found that a couple's probability of conception decreased for women over 30, and for women who smoked. We also found an association between decreased fertility and a women's exposure in utero to her mother's smoking. This association is being explored further in other data sets.